1. Field of the Invention
This invention relates to an apparatus and method for forming a layer of mercury cadmium telluride, and, more particularly, to such a method using an improved liquid phase epitaxy chamber.
2. Description of the Prior Art
Mercury cadmium telluride (Hg.sub.1-x Cd.sub.x Te) is typically used in the manufacture of infrared (IR) detectors, with commecial applications such as medical, thermography and building heat loss analysis. Detectors operating in the atmospheric windows are able to see both in the dark and through clouds. Hg.sub.1-x Cd.sub.x Te is the material preferred for use in the 3-5 um and the 8-14 um wavelength IR detectors.
Methods of Hg.sub.1-x Cd.sub.x Te crystal growth generally fall into three basic classes: growth from the melt, growth from solution, and growth from the vapor phase. All three of these methods and variations thereof have been, and are being used to grow Hg.sub.1-x Cd.sub.x Te, as described, for example, by J. L. Schmit, in the publication entitled "Growth, Properties and Applications of Hg.sub.1-x Cd.sub.x Te", in the Journal of Crystal Growth, Vol. 65, pages 249-261 (1983).
Liquid phase epitaxial (LPE) growth of Hg.sub.1-x Cd.sub.x Te films utilizing epitaxial chambers, however, has become the most often chosen means by those requiring efficient and cost effective production of Hg.sub.1-x Cd.sub.x Te.
The epitaxial boat design directly effects the growth of the film. Several factors present control limitations in current production operations. One factor is the difficulty in determining whether or not the substrate surface is exactly horizontal during growth in the tiltable (horizontal) growth chamber. A second factor involves the seal between the boat cover and the boat. Still another factor relates to the efficiency, quality, and cost effectiveness of quench anneal and doping processes.
As mentioned above, maintaining the substrate surface exactly horizontal during the Hg.sub.1-x Cd.sub.x Te crystallization/deposition onto the wafer substrate material is critical. For example, if the surface of the wafer is not exactly horizontal during growtn, the Hg.sub.1-x Cd.sub.x Te melt will be deeper over one portion of the wafer than another. This difference in melt depth will effect both the thickness and composition of Hg.sub.1-x Cd.sub.x Te crystallization on the wafer due to unequal depletions of mercury from both the surface of the melt, and from within the melt itself.
Heretofore, the usual check on proper horizontal positioning is by the operator's estimation of how uniformly the liquid melt is spread over the wafer surface after being positioned thereon--a purely qualitative estimate.
One prior method of determining the level of the boat required the introduction of a bubble level into the epitaxial chamber and properly leveling the epitaxial boat, the removal of the boat from the epitaxial furnace, the removal of the level and finally the repositioning of the boat. A typical liquid bubble level can be placed in the epitaxial boat while at room temperature to determine the level and adjustments made to effect exact horizontal positioning, but this bubble level must be removed and the boat repositioned exactly if this method is to work. Thus, a necessary sequential step is the removal of the bubble level without disturbing the boat, or removing the boat to remove the level and then properly re-positioning the boat in the exact original position. Once the boat is in the furnace, there is no way to positively determine whether or not the boat is still level. This practice is particularly limiting in as much as once the Hg.sub.1-x Cd.sub.x Te growth begins there is no absolute way to check the boat level before the termination of the growth, i.e., by cooling down and re-introducing the bubble level into the boat. This practice inevitably leads to the contamination of the growth materials due to excess handling of the boat and introduction of foreign contaminants.
Other prior methods of leveling utilize a single ball during a preliminary room temperature determination. This technique is inaccurate because this type of level detector must also be removed before the actual growth run and is unable to simultaneously and accurately determine a level on both the horizontal axes since it rolls randomly on the flat surface of the growth chamber.
Modified LPE methods that have markedly increased the liquid melt depth over the wafer to alleviate the leveling problem increase the cost of the epitaxial process and also introduce convection problems. These convection problems inherently effect the quality and uniformity of the Hg.sub.1-x Cd.sub.x Te crystal growth on the wafer.
A second above-mentioned factor directly affecting current LPE is the epitaxial boat sealing means. It has previously been a concern that despite the best efforts to seal a cover (normally constructed from quartz or similar inert material) over the epitaxial boat during the growth process, that mercury vapor would be lost from the chamber. Heretofore, the seal between the boat cover and the epitaxial boat has not been made impermeable by such gases. This was due to the necessity of the performance of an initial gas flushing step. This flushing procedure requires an efficient and complete replacement of the contained air which is normally introduced during the melt ingredient loading step. The requirements for this type of seal presented a dilemma, namely, to minimize mercury losses, the seal must be tight, but to permit good replacement of the atmosphere with an inert or reducing ambient, the seal must not be gas tight.
Heretofore, the sealing dilemma was attacked by costly epitaxial tuning runs. These runs were performed in an attempt to obtain the best possible seal tightness which allowed minimum escape of mercury and maximum atmosphere flushing capacity. These tuning runs, however, inevitably yielded non-uniform Hg.sub.1-x Cd.sub.x Te film growths from LPE growtn to LPE growth. Despite the best efforts to control the torque on the LPE boat seals, mercury continued to be lost during the epitaxial growth. Furthermore, the quartz LPE boat covers were thin and subject to bowing during the thermal stresses of epitaxial growth, thereby, allowing additional and non-reproducible mercury loss during epitaxial growth.
The third factor mentioned above pertains to the efficiency, quality and cost effectiveness of the quench anneal/doping process during LPE. Heretofore, the current practice is to produce a Hg.sub.1-x Cd.sub.x Te deposit on a cadmium telluride substrate (or on a cadmium telluride-coated alternate substrate, like sapphire) by liquid phase epitaxy. After a cool down of the epitaxy system, the sealed growth chamber is opened and the Hg.sub.1-x Cd.sub.x Te deposited wafer is removed and optically and physically characterized. A small test part can be measured by the Hall effect to determine some of the electrical properties which are critical to the deposited wafers final use in device fabrication. Typical electrical carrier concentrations, measured using the Hall effect at 77 Kelvin, are usually in the range of 10.sup.16 -10.sup.17 cm.sup.-3 ; however, a lower value in the 10.sup.15 range is more desirable for device material.
Accordingly, the lower value is achieved by an additional step in Hg.sub.1-x Cd.sub.x Te processing; namely, by annealing. Depending on the desired properties, the anneal/doping procedure can vary. One typical technique uses mercury vapor in the 300.degree.-400.degree. C. temperature range. This step currently may be accomplished by sealing the wafer bearing the Hg.sub.1-x Cd.sub.x Te film in a quartz capsule together with a small amount of mercury. Before seal off, the system must be pumped to a high vacuum to eliminate all gases other than mercury from the capsule. The sealed capsule is then inserted into a horizontal tube furnace and held at a predetermined anneal temperature, usually for a period of several hours, to allow mercury loss from the Hg.sub.1-x Cd.sub.x Te during epitaxial cool down to be restored into the Hg.sub.1-x Cd.sub.x Te film from the free mercury vapor. After the anneal, the capsule is quickly cooled so that the native defect concentration (see literature, for example: FIG. 20 on page 259 of J. L. Schmit's "Growth, Properties and Applications of Hg.sub.1-x Cd.sub.x Te", J. Cryst. Growth, Vol. 65, 249-261 (1983)) (this defect concentration usually being responsible for setting the measured carrier concentration value) now can be "frozen" into the material. Finally, the sealed quartz capsule must be broken open and the annealled wafer removed. These steps are, for a manufacturing operation, both time consuming and costly.
From the foregoing, it should be appreciated that present liquid phase epitaxy methods involving the deposition of Hg.sub.1-x Cd.sub.x Te are extremely factor dependent. Moreover, the multitude of variables involved in the proceoss makes the technique more of an art than a science, thereby resulting in a variable or non-uniform deposition of Hg.sub.1-x Cd.sub.x Te on wafer substrates.
Accordingly, a fuller understanding of the invention may be obtained by referring to the Summary of the Invention, and the Detailed Description of the Preferred Embodiment, in addition to the scope of the invention defined by the claims taken in conjunction with the accompanying drawings.